| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Rous-Whipple Award Lecture |
From the Laboratory of Molecular Pathology, Department of Pathology, The University of Texas Southwestern Medical Center, Dallas, Texas
"I do not know what I may appear to the world, but to myself I seem to have been only like a boy playing on the sea-shore, and diverting myself in now and then finding a smoother pebble or a prettier shell than ordinary, whilst the great ocean of truth lay all undiscovered before me."
Isaac Newton
The genomes of all living cells are constant targets for
DNA damage.1
Alterations of the bases in DNA can arise
spontaneously as a result of errors during DNA replication, or due to
alterations in the chemistry of the DNA bases. Such damage can also
result from exposure of cells to a host of reactive environmental
physical or chemical agents, such as ultraviolet (UV) radiation or a
plethora of mutagenic and carcinogenic agents.1
All
free-living life forms have evolved multiple distinct responses
to DNA damage and other insults to the function of the genome that can
be collectively considered as biological responses to DNA damage
(Figure 1)
. Signals that are initiated by
various types of DNA damage are transduced through a number of complex
damage checkpoint pathways, resulting in arrested cell cycle
progression at various stages. This allows for increased time for
repair of the DNA damage before continuing progression through the
cycle. When cells are faced with arrested DNA replication due to the
presence of base damage or DNA strand breaks, altered forms of DNA
synthesis occur that essentially tolerate the presence of the base
damage without its physical removal. The altered DNA synthesis can be
either error-free or error-prone. In the latter instance mutations are
generated. Finally, one of the primary ways in which living cells
respond to DNA damage involves a series of biochemical reactions that
are collectively referred to as DNA repair.1
My own
interest in this group of biochemical pathways began over 30 years ago
as a postdoctoral fellow in the Department of Biochemistry at Case
Western Reserve University, and was strongly reinforced when I
encountered James E. Cleaver’s description of defective
DNA repair in cells from individuals
suffering from the skin cancer-prone hereditary disease xeroderma
pigmentosum (XP).2
|
The phenomenon of excision repair of DNA damage associated with
exposure to ultraviolet (UV) radiation was discovered in bacteria by
Richard Setlow and his colleagues in the mid-1960s.3
Nothing was known about excision repair in mammalian cells. At about
this time Robert Painter and his associates developed a simple
autoradiographic technique for monitoring DNA synthesis in mammalian
cells to study the effects of ionizing radiation on DNA replication. To
their surprise, they observed that when cells were exposed to UV
radiation, low levels of DNA synthesis were also observed in cells that
were not actively engaged in DNA replication, ie, cells that were
outside the S phase of the cell cycle (Figure 2)
. In light of the discovery of excision
repair in bacteria, Painter and his colleagues speculated that the low
levels of DNA synthesis in UV-irradiated mammalian cells might reflect
repair synthesis specifically associated with excision
repair.4
Shortly after this observation, Cleaver joined
Painter’s laboratory. To validate Painter’s experiments with an
appropriate negative control as Setlow and his colleagues had done by
using an excision repair-defective mutant strain of Escherichia
coli, Cleaver was preparing to mutagenize mammalian cells to
generate such a mutant, when he encountered a description in the San
Francisco Chronicle of the high incidence of skin cancer in
XP patients exposed to sunlight. He reasoned, correctly, that XP
individuals might be skin cancer-prone because of a defect in the
putative excision repair mode observed by autoradiography in normal
cells. If so, XP cells would validate the autoradiographic
demonstration of excision repair in normal human cells. This experiment
led to the demonstration of defective excision repair in XP cells in
culture.2
|
XP-G.1
These observations, in
turn, suggested that multiple genes are required for the particular
mode of excision repair demonstrated in normal human cells and that
mutational inactivation of any one of these genes results in XP. By this time it was also established that there are three biochemically distinct modes of excision repair in nature.1,6 These are designated as nucleotide excision repair (NER), a process required for the repair of bulky base adducts in DNA such as those resulting from exposure to UV radiation; base excision repair (BER), a totally distinct biochemical pathway required for the repair of many types of non-bulky base damage such as that which occurs spontaneously from reactive oxygen species in cells; and mismatch repair (MMR), a process required for the excision of mispaired bases in DNA.
XP individuals are specifically defective in NER.2,7
As a
consequence their cells are highly sensitive to killing by UV
radiation and to many chemical mutagens and carcinogens that result in
bulky base damage in DNA1
(Figure 3)
. The existence of multiple genetic
complementation groups for the disease strongly suggested a profound
biochemical complexity for NER in normal human cells. With the advent
of the era of recombinant DNA technology, the marked difference in the
sensitivity of normal and XP cells to UV radiation invited a gene
cloning strategy based on screening the genome from normal human cells
for phenotypic correction of UV radiation sensitivity. Several
investigators embarked on this heroic experimental approach, despite
the fact that in the decade of the 1970s the cloning of human genes by
phenotypic complementation using human cells in culture was a
formidable technical challenge.
|
Dissuaded by the technological complexities of human gene cloning
at that time, my colleagues and I began to investigate the yeast
Saccharomyces cerevisiae as a eukaryotic model for defective
NER.8
Yeast mutant strains existed that, like human XP
cells, were exceptionally sensitive to killing after exposure to
UV radiation or selected chemicals. These were designated as
rad mutants (for
radiation-sensitive).1
Multiple distinct
rad mutants were known to exist, indicating a level of
genetic complexity for NER in yeast comparable to that suggested by the
multiple XP genetic complementation groups in human cells. In the late
1970s and early 1980s, we and others characterized the DNA repair
phenotype of various yeast rad mutant strains and
established that they are indeed defective in NER in
vivo.9,10
Encouraged by these observations, we
embarked on the molecular cloning of yeast genes by phenotypic
complementation of UV radiation sensitivity, which was technically much
more facile than cloning human genes (Figure 4)
. In short order we and others cloned
and characterized a series of genes now designated RAD1, RAD2,
RAD3, RAD4, RAD7, RAD10, RAD14, RAD16, and RAD23, all
of which were shown to be indispensable for the process of NER in
yeast.11,12
|
The genetic and biochemical complexity of NER in yeast has now
increased to at least 19 proteins (Table 1)
, all of
which are required for events that precede repair synthesis and DNA
ligation. Overexpression of these genes in bacterial cells provided the
starting material for the systematic purification and characterization
of individual polypeptides required for NER. These studies have
culminated in the identification of specific catalytic functions
associated with particular polypeptides or polypeptide complexes.
Additionally, highly conserved homologous genes have now been cloned
from human cells and their polypeptide products have been purified
(Table 1)
. Many of these genes have indeed turned out to be mutated in
XP individuals. Reconstituted human and yeast systems that support NER
in vitro have now been developed with recombinant
proteins.16-19
|
, even in the
absence of exposure of cells to DNA damage of the type that requires
NER. This repairosome (or NER machine) may patrol the genome in a
surveillance mode, and the recognition of appropriate substrates for
NER may be a function of multiple interacting components of the entire
complex rather than any single polypeptide. A definitive answer to the
question of how cells recognize so many chemically distinct forms of
base damage during NER still eludes us.
|
. This denaturation is believed to be
catalyzed by two of the TFIIH subunits, Rad3 (human XPD) and Ssl2
(human XPB) proteins, both of which are DNA helicases with opposite
directionality31,32
(Figure 6A)
. A similar, if not
identical function of the Rad3 (human XPD) and Ssl2 (human XPB)
helicases is subserved during RNAP II transcription (Figure 6A)
. By
this process, limited melting of DNA is achieved, which facilitates the
initiation of mRNA synthesis and the clearance of the promoter region
by the transcription apparatus.16-19
Though apparently
co-opted for the process of NER, localized unwinding of DNA during this
process serves an entirely different purpose. The products of the
RAD1 (human XPF) and RAD10 (human
ERCC1) genes function in a heterodimeric complex which
specifically endonucleolytically cleaves the junction between
single-stranded and duplex DNA in substrates in which the
single-stranded DNA has 3' directionality33
(Figure 6B)
.
Correspondingly, the Rad2 (human XPG) protein is a DNA
junction-specific endonuclease that cleaves between single-stranded and
duplex DNA in substrates in which the single-stranded DNA has 5'
directionality34
(Figure 6B)
. Cleavage of the
single-stranded/double-stranded junctions generated at each end of the
localized denaturation bubble will, if confined to the DNA strand
carrying the damaged base, generate incisions (nicks) that flank the
damaged base (Figure 6B)
. It is not known how cleavage of the opposite
undamaged strand is avoided. A reasonable guess is that the topology of
the repairosome-damaged DNA complex is such that the catalytic sites of
the Rad1/Rad10 (human XPF/ERCC1) and Rad2 (human XPG) endonucleases are
placed in immediate proximity to the duplex/single-stranded DNA
junctions of only the damaged strand. Alternatively or additionally,
the topological organization of the repairosome-damaged DNA complex may
provide protection of the undamaged strand by single strand binding
protein (SSB), which is known to be required for normal NER and is an
integral component of the repairosome.
|
The dual function of TFIIH in both RNAP II transcription and NER raises the interesting evolutionary question of how this came about. A reasonable speculation is that an early eukaryotic form of the NER apparatus, which may now be extinct, did not use TFIIH and operated in a biochemically distinct mode, perhaps more akin to that known in E. coli and other prokaryotes.1 However, a spurious binding affinity between this NER complex and one or more subunits of TFIIH promoted the sequestration of this complex from the RNAP II transcription machinery and resulted in reduced efficiency of transcription initiation. This phenotypic consequence was selected during evolution because reduced transcription in the presence of DNA damage afforded a significant protective mechanism against the mutagenic and/or lethal effects of genotoxic agents. In time this selection evolved to the permanent physical incorporation of TFIIH into the NER complex, and this in turn promoted a new mechanism for damage-specific incision of DNA, which is predicated on the DNA helicase function of TFIIH. Regardless of the accuracy of this speculation, the fact that TFIIH functions in both RNAP II transcription and NER affords potential regulation of each of these processes in the presence of the other. Indeed, work in my laboratory has directly demonstrated that RNAP II transcription of a yeast gene in vitro is inhibited in the presence of active NER.35
The dual incisions that flank sites of base damage generate
oligonucleotides 
30 nucleotides in length that are no longer
covalently attached to DNA. The physical excision of these
oligonucleotides requires further components of the yeast repairosome.
Specifically, proteins called Rad7, Rad16, and Abf1 form a tight
complex and are believed to act in concert to facilitate
oligonucleotide fragment excision36
(Figure 6C)
. The
precise function of the Rad7/Rad16/Abf1 repairosome subcomplex is not
known. Whereas most of the proteins involved in NER in yeast are well
conserved in humans, obvious homologues of the yeast RAD7,
RAD16, and ABF1 genes have not been detected in
the human genome and are not present in the genome of Drosophila
melanogaster, the sequence of which was recently completed. Hence,
the mechanism by which damage-containing oligonucleotide fragments are
excised may differ in lower and higher eukaryotes. Regardless, one
anticipates that the process of oligonucleotide excision is coupled to
repair synthesis of DNA in both yeast and human cells. In this way
nucleotides can be replaced in the genome essentially concurrently with
their removal during NER (Figure 6C)
and the formal existence of large
single-stranded gaps in the genome can be avoided. When repair
synthesis is accomplished, DNA ligation restores the complete
covalent integrity of the genome and the process of NER is completed.
This brief summary of the mechanism of NER completely has deliberately omitted consideration of the fact that in eukaryotic cells DNA is structurally organized into chromatin. The vexing question of how this intimate association of DNA with numerous histone and non-histone chromosomal proteins is disrupted and reconstituted to allow access of the NER machinery to sites of base damage remains a challenging issue.
The Return to Mammals: Mice as Model Organisms
The essential features of the biochemistry and molecular biology of NER in eukaryotes are now well understood. However, many questions concerning the relationship between defective NER and cancer predisposition in the human disease XP remain to be answered. XP is a rare disease; hence, the availability of human subjects for study is limited. Additionally, one is faced with the many limitations associated with all human experimentation and, not inconsequentially, the heterogeneity of genetic and other factors (such as age) that is intrinsic to human populations. Because (regrettably, from an experimental point of view) yeast do not develop cancer in the sense that we understand this disease in multicellular organisms, in the mid-1990s I sought a new experimental model that was free of these limitations, to begin to investigate the relationships between defective NER and cancer predisposition, and elected to join the new wave of investigators exploiting the many uses of genetically engineered mice.
Using conventional targeted gene replacement, my colleagues and I inactivated the mouse Xpc gene, the highly conserved (>90% amino acid identity) orthologue of the human XPC gene.37-39 Human patients from the XP-C genetic complementation group have the same primary clinical features of extreme photosensitivity and skin cancer predisposition observed in other XP individuals.1,40 However, at the molecular level they uniquely retain proficiency for NER of the transcribed strand of transcriptionally active genes, whereas other XP patients are totally defective in NER.1 The only clinical phenotypic association that has been related to this difference is a reduced incidence of neurological dysfunction in XP group C patients compared to other XP individuals.40
Like their human counterparts,
Xpc-/- mice are
defective in NER of transcriptionally silent DNA and that of the
nontranscribed strand of transcriptionally active genes, but retain the
ability to repair UV radiation photoproducts in the transcribed strand
of transcriptionally active genes.37
The mice suffer a
high predisposition to skin cancer after exposure to UVB
radiation41
(Figure 7)
. By
25 weeks after the onset of a regimen of daily exposure to UV light
for 18 weeks, 100% of these animals develop cancers on the shaved
dorsal skin (Figure 8)
. No skin cancers
are observed in Xpc+/- and
Xpc+/+ littermates at this time (Figure 8)
.
However, by about 70 weeks after initiation of the radiation protocol,
Xpc+/- animals are clearly more
cancer-prone than wild-type controls41
(Figure 8)
. This
result apparently reflects true haploinsufficiency at the
Xpc locus, since no mutations have been detected in the
remaining Xpc allele in a number of tumors
examined.41
Extrapolating to humans, this observation
raises the interesting and important possibility that individuals who
are heterozygous for the XPC mutation may have an increased
risk of skin cancer associated with sunlight exposure.
|
|
One of the unique opportunities afforded by the availability of increasing numbers of genetically engineered mice is the potential for crossing different mutant strains to generate strains with multiple mutations in relevant genes. Among numerous such experiments of this type, my colleagues and I generated all possible combinations of Xpc and Trp53 (the mouse name for p53) heterozygous and homozygous double mutant mice. When Xpc-/- mice missing one or both copies of the Trp53 gene were exposed to UVB radiation, skin cancers appeared significantly earlier in a gene dosage-dependent manner.45 Aside from its known functions in cell cycle checkpoint control and apoptosis, human p53 protein has been directly implicated in NER.46,47 Hence, it is conceivable that the increased predisposition to skin cancer in Trp53 mutant mice reflects inactivation of the residual proficiency of Xpc cells for repair of the transcribed strand of transcriptionally active genes. Alternatively, and more likely in my view, the increased predisposition to skin cancer in Trp53 mutant mice reflects the additive effects of defective NER and defective p53 function as a tumor suppressor gene.
Because the Trp53-null phenotype associated with loss of both alleles increases the rate of development of skin cancer more dramatically than the Trp53+/- genotype, we reasoned that the later appearance of skin cancers in the latter genetic background reflects the time taken to inactivate the remaining Trp53 allele due to a failure to repair UV radiation damage. We therefore sequenced the remaining Trp53 allele in about 20 skin cancers in Xpc-/- Trp53+/- mice and confirmed mutations in >90% of the tumors.48 Most unexpectedly, in addition to the mutational hot spots observed by us and others in the Trp53 gene in wild-type animals, all of which are at dipyrimidine sites where major photoproducts (pyrimidine dimers and [6–4] photoproducts) are known to occur, we observed an unusually high frequency of mutations affecting the C residue in the sequence ACG, the coding trinucleotide of codon 122 of the p53 open reading frame.48 In this sequence C is flanked by purines; hence, pyrimidine dimers and [6–4] photoproducts cannot be generated. This C residue is believed to be methylated, as it is part of a CpG dinucleotide.
The observation of a UV radiation-dependent mutational hot spot in the Trp53 gene raises several interesting questions. What is the nature of the putative photoproduct formed at this non-dipyrimidine CpG site? Does this represent a rare photoproduct that is normally repaired by the NER pathway, or conceivably by a novel repair pathway that requires the XPC protein but not other NER proteins? In an effort to address the latter issue we examined the mutational spectrum in the single Trp53 allele in skin cancers generated by UVB irradiation of Xpa-/- Trp53+/- mice. No mutations affecting codon 122 were detected in 11 skin cancers examined (D Nahari, LB Meira, and EC Friedberg, unpublished observations). In a twelfth tumor, an inactivating mutation was located elsewhere in the Trp53 gene, and a single second mutation in the same tumor was found in codon 122 (D Nahari, LB Meira, and EC Friedberg, unpublished observations). These preliminary experiments suggest that the high frequency of mutations in codon 122 is indeed Xpc gene-specific. Whether this specificity relates to differences between Xpc and other XP mutant mice in the biology of neoplastic transformation associated with exposure to UVB radiation, or to differences in the repair of a rare photolesion in DNA, remains to be established and is the subject of ongoing studies.
Other observations suggest the possibility that XPC protein may be specifically required for the repair of a rare form of base damage associated with exposure to UVB radiation. The frequency of spontaneous mutations measured in the Hprt gene of lymphocytes increases dramatically (20- to 30-fold) as a function of age in Xpc-/- mice, but not in Xpa-/- mice (SWP Wijnhoven, HJM Kool, LHF Mullenders, AA van Zeeland, EC Friedberg, GTJ van der Horst, H van Steeg, H Vrieling, unpublished observations). These mutations presumably arise from unrepaired spontaneous base damage in DNA, such as that generated by reactive oxygen species in cells.
Trp53-null mice spontaneously develop various types of malignant tumors.49 Initially these are predominantly testicular teratocarcinomas. However, lymphomas and other sarcomas are prominent later.49 However, of relevance to the present discussion, when such mice are also genotypically Xpc-/- the kinetics of the appearance of testicular (but not other) tumors is significantly increased.43 This observation also suggests that some form of spontaneous base damage in the DNA, this time in testicular cells, requires XPC protein for efficient repair.
Acknowledgements
This article is the gist of the Rous-Whipple Award Lecture presented at the Annual Meeting of the ASIP on April 18, 2000. I wish to thank the numerous graduate and medical students, postdoctoral fellows, and residents at Stanford University and The University of Texas Southwestern Medical Center, with whom I have had the good fortune and privilege to work for the past 30 years. Special thanks are due to John Feaver, Simon Reed, Paula Fischhaber, David Cheo, Lisiane Meira, Antonio Reis, and Dorit Nahari, present and recent members of my laboratory who contributed to the story told here, and to Valerie Gerlach and Lurdes Queimado for critical review of the manuscript. I also wish to thank my many peer collaborators around the world, in particular Roger Kornberg at Stanford University, for many exciting and stimulating collaborations. Finally, I thank my earliest mentor in serious science, David Goldthwait, for allowing a complete novice to enter his laboratory at Case Western Reserve University and for patiently guiding him in the right direction, and David Korn, former Chair of the Department of Pathology at Stanford University, for providing strong leadership and unflagging support to a fledgling independent investigator.
Footnotes
Address reprint requests to Errol C. Friedberg, M.D., University of Texas Southwestern Medical Center, Department of Pathology, Dallas, TX 75390-9072. E-mail: friedberg.errol{at}pathology.swmed.edu
Studies in the author’s laboratory are funded by research grants CA12424, CA44247, and CA69029 from the National Cancer Institute, National Institutes of Health.
Accepted for publication May 18, 2000.
References
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
or 
,
together with accessory proteins for DNA replication
(RPA, RFC, and PCNA) fill
in the gap generated by oligonucleotide excision. DNA ligation
completes NER.
